Re: Requirements for a new stereo camera

[bercov@xxxxxxxxxxx (John Bercovitz)]

First I would like to make small comments on specific numbers in 
the draft proposal for a new stereo camera.  Later I will post 
suggested specifications that fall within the bounds of the 
present proposal.  This because I think we need to take a more 
directive approach when dealing with the camera companies: give 
them solid numbers, a standard.  If we do the research then they 
don't have to do it.  Later let the market decide if stereo really 
must move away from the standard.

I think that matching focal lengths within 0.2% will be difficult 
to achieve at a reasonable price.  However, if you loosen this 
tolerance to 0.5%, will it not still give you at maximum an 0.1 mm 
error contribution?  

I think the autofocusser should find near and far distances and 
set the correct focus distance to match the amounts of out-of-
focus for near and far objects.  This shouldn't be too difficult 
for a chip and you've got to have a chip anyway in order to do all 
that is being proposed.  The biggest problems would be locating 
the nearest and farthest objects in the scene so as to find their 
distances, and having enough sensors to cover the possibilities.  
This will call for some design cleverness.

I think most of the people on this list would like to see aperture 
priority with depth of field scales as the primary autoexposure 
mode.  A program mode geared to set the depth of field for the 
near and far distances determined above would be a good addition.

I would set the stereo window at 1.5 m instead of 2 m.  I don't 
think many people would have trouble with the closer window, but I 
also think experiments should be done to see if this is true or 
not.  This way you could, in general, avoid the problems of so 
many masks.

John B



------------------------------